Identifying Youth at Risk for Problematic Absenteeism Using Nonparametric Modeling: The Impact of Youth Psychopathology and Family Environment Risk Factors
Abstract
The best cutoff to differentiate problematic school absenteeism from nonproblematic school absenteeism has yet to be identified in the literature (Lyon & Cotler, 2007). Contemporary classification systems, including Multi-Tiered Systems of Support (MTSS), depend upon cutoffs to clearly define the various tiers (Jimerson, Burns, & VanDerHeyden, 2016). The current study aimed to inform the MTSS approach while also contributing to early identification, assessment, and intervention methods for those youth and families at the highest risk of problematic school absenteeism and its negative consequences. The current study identified subgroups of youth at the highest risk of problematic absenteeism, defined as equal to or greater than 1% of full days missed and equal to or greater than 10% of full school days missed cutoffs (Egger et al., 2003; NCES, 2016). Interactions among family environment and youth psychopathology risk factors were evaluated at each cutoff. Participants included 378 elementary, middle, and high school students and their families from clinic and community settings. The current study utilized nonparametric Classification and Regression Tree (CART) procedures via SPSS decision tree software. CART’s procedures are meant for generating hypotheses and not testing hypotheses (Markham, Young, & Doran, 2013). Therefore, hypotheses provided were based on the extensive literature base of problematic school absenteeism risk factors. Hypothesis one was that Family Environment Scale (FES) items addressing family conflict were expected to be the most important FES items to the model while independence items were expected to be the second most important. Hypothesis two was that Revised Children’s Anxiety and Depression Scale (RCADS) items addressing generalized anxiety were expected to be the most important RCADS items to the model while major depression items were expected to be the second most important. Post-hoc analyses were also conducted to explore additional cutoff scores (i.e., <1%, 3%, and 5%), gender distinctions (i.e., male and female), and developmental distinctions (i.e., children and adolescents). Hypotheses were partially supported. Implications for clinicians, researchers, and educators are discussed.